Closed jmrohwer closed 1 month ago
Hi @jmrohwer, just finishing off the academic year, will dive into the review next week.
Quick note: I created a new Python10 environment to test the release in and decided to try using RTools 4.4. I followed the PySCeS INSTALL instructions and only had to change the paths to get it to work:
c:\rtools40\usr\bin
c:\rtools40\ucrt64\bin
to
c:\rtools44\usr\bin
c:\rtools44\x86_64-w64-mingw32.static.posix\bin
I haven't tested it extensively yet, and am not even sure if this is the correct way to do it, but it seems to work.
Quick note: I created a new Python10 environment to test the release in and decided to try using RTools 4.4. I followed the PySCeS INSTALL instructions and only had to change the paths to get it to work:
c:\rtools40\usr\bin c:\rtools40\ucrt64\bin
to
c:\rtools44\usr\bin c:\rtools44\x86_64-w64-mingw32.static.posix\bin
I haven't tested it extensively yet, and am not even sure if this is the correct way to do it, but it seems to work.
Yes, I've tested it as well and it works fine. However, it is a 428M download compared to 166M for version 4.0, and version 4.0 works equally well. It is also what numpy and scipy uses, so I've oriented myself by that.
Now that numpy 2.0 has been released it is time to cut another PySCeS release as all extension modules need to be recompiled against numpy 2.0 to work with
numpy>=2.0
. The minimum supported numpy version is now 1.23.5 in line with scipy and other downstream projects.This PR fixes some numpy 2.0 compatibility-related bugs and updates requirements.
In addition:
dill
to pickle model object).@bgoli Nothing seems controversial here but I would appreciate your quick glance. I've tested the module with both
numpy=2.0.0
andnumpy=1.23.5
and all CI builds pass (with their built in tests).